CRISP: Correlation-Refined Image Segmentation Process

Abstract

Calcium imaging enables real-time recording of cellular activity across various biological contexts. To assess the activity of individual cells, scientists typically manually outline the cells based on visual inspection. This manual cell masking introduces potential user error. To ameliorate this error, we developed the Correlation-Refined Image Segmentation Process (CRISP), a two-part automated algorithm designed to both enhance the accuracy of user-drawn cell masks and to automatically identify cell masks. We developed and tested CRISP on calcium images of densely packed {beta}-cells within the islet of Langerhans. Because these {beta}-cells are densely packed within the islet, traditional clustering-based image segmentation methods struggle to identify individual cell outlines. Using {beta}-cells isolated from two different mouse phenotypes and imaged on two different confocal microscopes, we show that CRISP is generalizable and accurate. To test the benefit of using CRISP in functional biological analyses, we show that CRISP improves accuracy of functional network analysis and utilize CRISP to characterize the distribution of {beta}-cell size.